Branched equation modeling of simultaneous accelerometry and heart rate monitoring improves estimate of directly measured physical activity energy expenditure
Brage, 2003
Last update: January 15, 2022. Updates will occur every 3 months.
Branched equation modeling of simultaneous accelerometry and heart rate monitoring improves estimate of directly measured physical activity energy expenditure
Brage, 2003
Models to predict energy expenditure from CSA accelerometer counts/min and/or heart rate were created in a sample of 12 males using individual and group calibration approaches. Percent error for each model was: CSA only (individual calibration: -50.8, group calibration: -45.1), heart rate only (individual calibration: 39.1, group calibration: 48.8), non-branched CSA plus heart rate (individual calibration: 29.9, group calibration: 25.7), branched CSA plus heart rate using a priori parameters, (individual calibration: -4.4, group calibration: 3.5) branched CSA plus heart rate using post hoc parameters (individual calibration: -1.5, group calibration: 0.1).
Sample: 12 males, 22.7-30.0 years of age
Setting: Laboratory
Activities: Activities of daily living, cycling, stationary, stepping, treadmill walking and running
Criterion: EO Sprint by Jaeger (VO2), heat sink calorimeter
Accelerometer(s): CSA on each hip
Validation approach: Comparison to criterion
Multiple models were created- CSA only, heart rate only, non-branched CSA plus heart rate, branched CSA plus heart rate with parameters determined a priori, branched CSA plus heart rate with parameters determined post hoc, each using group and individual calibration. Physical activity intensity (PAI) is reported in kilojoules per kilogram per minute by and defined as VO2 minus measured RMR. Heart rate is expressed as beats/minute above resting heart rate. CSA is the mean accelerometer values per minute.
The flex point was defined as 50% of the mean CSA output at 3 km/h. At the group level, this value was 497 counts/min. Above the flex point, PAI is predicted from the equation 0.053*CSA + 47.88. Below the flex point, PAI is calculated as 0.15*CSA.
The flex heart rate was defined as 10 beats/min plus the average of resting heart rate and the mean heart rate at 3 km/h. Below this flex heart rate, PAI is 0 kJ/kg/min. Above flex heart rate, PAI is predicted from the equation 0.011*heart rate2 + 5.82*heart rate.
PAI is predicted using the equation 0.028*CSA + 4.04*heart rate
Branched models follow the following flow chart and used either a priori parameters or post hoc parameters which are found below.
y= walking/running transition heart rate (mean heart rate between fastest walking and slowest running speed)
z= flex heart rate (10 beats/min plus the average of resting heart rate and the mean heart rate at 3 km/h)
P1=100%
P2=P3=50%
P4=0%
x=5 counts/min
Parameters are found in the table below.
Reference:
Brage, S., Brage, N., Franks, P. W., Ekelund, U., Wong, M. Y., Andersen, L. B., ... & Wareham, N. J. (2004). Branched equation modeling of simultaneous accelerometry and heart rate monitoring improves estimate of directly measured physical activity energy expenditure. Journal of Applied Physiology. https://doi.org/10.1152/japplphysiol.00703.2003 Link to Paper
Corresponding author: Søren Brage, sb400@medschl.cam.ac.uk
Kimberly Clevenger at accelerometerrepository@gmail.com